- User Since
- Dec 27 2014, 8:52 PM (317 w, 3 d)
@ldionne: I don't think buildkite is ever going to finish a build here. I propose that you approve this revision and I just go again and land it, and we'll see what happens with the buildbots after that. (If it somehow does break buildbots, we can revert it.) Thoughts?
Poke buildkite yet again. One day it'll finish a build.
Poke buildkite yet again. ( https://reviews.llvm.org/B86662 failed in the same way.)
Mon, Jan 25
Poke buildbot. (The last build https://reviews.llvm.org/B86646 seems to have failed in a devopsy way, and clicking "Restart Builds" just tells me I haven't got permissions for that.)
Add "simple" ad-hoc tests at @ldionne's request.
Poke the buildkite.
Silence -Wignored-qualifiers also for GCC. If this makes CI green, I'll land this.
Sun, Jan 24
Add some unit tests for pointers-to-member(-function)s, as suggested by @logan-5.
Address review comments and try to silence -Wignored-qualifiers warnings.
Fri, Jan 22
Thu, Jan 21
The theme of the answers to many of your questions is "because the standard library sucks on purpose." Both is_convertible and declval are unfit for purpose in most places where a metaprogrammer might think about using them.
Support std::function<const NonCopyable()>, and then discover that std::function<const void()> wasn't fully supported either. This is now fixed.
However, the unit test for the fix might be unacceptable, because it inevitably triggers -Wignored-qualifiers and I can't figure out how to shut it up. @ldionne, any ideas?
Tue, Jan 19
https://isocpp.org/std/standing-documents/sd-6-sg10-feature-test-recommendations seems out of date anyway, the draft standard is maybe a better source? https://eel.is/c++draft/version.syn
Mon, Jan 18
@ldionne: LGTM except that the non-constexpr tests no longer need to return true; from bool test() — they could do void test() for now.
Also, I don't think this will rebase as-is on generate_feature_test_macro_components.py, will it? You'll have to add the entry for __cpp_lib_string_contains, which will be easy — but I notice that it will be a divergence from what's literally in https://isocpp.org/std/standing-documents/sd-6-sg10-feature-test-recommendations as of this writing. I would bet that's fine; I'm just bringing it up for awareness.
Sat, Jan 16
Rebase, and fix my new test's constexprness in C++11 mode.
Fri, Jan 15
In the new test, make TCBNTM and TMBNTC copy-constructible, because if they're not, then you can't make an array of them pre-C++17. ("error: copying array element of type 'TMBNTC'...")
Tests failed. Let's try this again.
Improve the contiguous_iterators.pass.cpp test, including a test case for exactly what I'd expect Chromium to write in their code to enable contiguous-iterator-ness.
Wed, Jan 13
Tue, Jan 12
@nullptr.cpp: I may or may not come up with more review comments, but either way I'm a dead end — I don't consider myself authorized to "accept" Clang changes.
You're still going to have to attract the attention of someone with the authority and/or force-of-will to say "okay ship it."
Seems like a generally good direction to me, FWIW.
Mon, Jan 11
Merged some test cases from D60027.
Rebased on main. @ldionne ping! :)
Fri, Jan 8
Rebased on main. @ldionne ping!
Whoops, my rebasing lost the UNSUPPORTED entries for barrier/latch/semaphore. Restored.
Thu, Jan 7
Push the alphabetization part (thanks @ldionne!), and rebase.
Rebased on main after D93383.
Wed, Jan 6
Tue, Jan 5
LGTM, but redundant with D93830.
Mon, Jan 4
Sun, Jan 3
Fri, Jan 1
This could use some tests. (What's observably bad about the old code's behavior? does this patch fix that behavior?)
Wed, Dec 30
Add ["UNSUPPORTED: libcpp-has-no-threads"] to barrier, latch and semaphore.
Tue, Dec 29
Dec 28 2020
Fix a failing test (don't put constexpr on non-const methods in C++11).
If buildkite is green after this, imma push it.
Dec 27 2020
Poke the buildbot. (It's been stuck in "pre-merge checks" for 10 hours.)
Dec 26 2020
Run constexpr tests also in non-constexpr mode. Rebase on main.
Dec 25 2020
Complete the implementation of __cpp_lib_constexpr_functional and set that feature macro. This involves constexprifying std::default_searcher from P1032 "Misc constexpr bits". There's more "misc" stuff in P1032, which I have not attempted to do in this PR because it doesn't affect the __cpp_lib_constexpr_functional feature macro.
Dec 24 2020
I don't fully understand the new control flow, but at least the new behavior (after applying this patch) looks like an improvement to me.
I recommend rebasing on top-of-tree, mainly so that the buildbots will run again and presumably be greener this time.
Dec 23 2020
Dec 22 2020
Rebase in preparation for finally pushing this thing if the buildbots seem happy.
Rename "initial_header" to "initial_descriptor" (since technically it's neither a header nor a footer).
Rebase on main.
Dec 21 2020
Add feature-macro stuff and update the CSV files.
Dec 20 2020
Dec 18 2020
Dec 17 2020
The buildbot was indeed happy. Remove the *_heap algorithms (in preparation for their separate PR). This is now ready for landing AFAIC.
Dec 16 2020
Dec 14 2020
Dec 11 2020
The GCC test failure is interesting. I just filed https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98249 about it, and will wait to see what develops.
No change since last time. Rebase on D93137, and poke buildbots.
Rebase, to poke the buildbot.
Dec 10 2020
Rebased on D92875 for the benefit of buildkite. If buildkite is green, I'm going to land both of these.